As the Corona Virus (SARS-CoV-2) continues to spread globally, causing severe human disease and death (COVID-19), citizens in Switzerland and around the world have been exposed to a range of prevention, containment and control measures, communicated by a diverse spectrum of media outlets, including information channels of governments and public health authorities but also newspapers, radio and television broadcasts, and social media platforms. During this public health emergency, official policy briefings, expert opinions, commentaries, idiosyncratic views, popular sentiments as well as dashboards, interactive maps and visuals have not only become a source of information, but also an incubator for misunderstandings, emotional responses, moral judgements, and behavioral changes in daily routines. Clearly, not all communication content is equally reliable, and citizens vary in the information sources they can or choose to access. Providing high-quality information while actively dismantling myths is a key concern for national and global health authorities, as exemplified by the Swiss Federal Office of Public Health (FOPH) and the World Health Organization (WHO) information campaigns. These efforts, however, need to be complemented by endeavors to obtain an excellent grasp of public perception in order to allow for continuous adaptations and improvement of emergency communication as well as prevention and containment strategies. Failure to communicate effectively might result in irritation, loss of trust, and suboptimal adherence to public health policies.
Against this backdrop, our project aims to develop a tool that helps tackle the “infodemic” manifested in the COVID-19 context, with a focus on a nuanced and in-depth understanding of public perception. With this tool, we intend to foster effective and tailored risk and crisis communication as well as an assessment of the risks and benefits of prevention, containment and control measures, as their effectiveness will crucially depend on public trust and cooperation.
The project adopts a trans-disciplinary multi-stakeholder approach including a participatory citizen science component. Methodologically, we propose a combination of literature and media review and analysis, and empirical studies using mixed methods. Building on real-time data and continuous data collection, our research results will be highly adaptable to the evolution of the current emergency (e.g., with respect to vaccine development). We expect the tool to be readily translatable not only to other health systems but also to other public crises, such as disaster management, refugee crises, and other public safety and civil protection scenarios, where national and international authorities will need to grapple with the development of crisis communication and intervention strategies to address public needs and interests.
COVID-19, infodemiology, public health interventions, population behaviour, disaster response, preparedness, participatory science
Swiss National Science Foundation, project number 31CA30_195905
2020 - 2022
Firstly, the project will be informed by an ongoing review and analysis of the existing and emerging literature and of other communication tools and platforms. The literature review will be complemented by a media analysis. While it is outside the scope of this project to provide a comprehensive content analysis of the media landscape on COVID-19, we will study media events that significantly influence public understanding, emotional state, or behavioural dispositions as revealed by PubliCo survey data. The analysis will not be limited to the so-called quality media but draw on information sources that are cited by survey respondents. We will focus on written text while expanding the analysis to audio-visual material. We will use Factiva as a database for media communication, as well as various media webpages and social media platforms, including Twitter, Facebook and YouTube (comment section). The sampling strategy will be continuously informed by the results of the online platform.
The online platform designed for data collection will be developed in cooperation with Belka, a software house based in Trento/Munich, with extensive expertise in user experience design and development. The tool will be web-based, mobile first, and will have an open structure, allowing non-technical staff to add and modify both survey items and information for the users. The content management system will support a multilingual interface. User experience testing will help ensure the survey is clear and accessible to a large part of the Swiss population. Data will be collected, stored and managed in accordance with Swiss law. Quantitative data will be collected anonymously. Sociodemographic variables (e.g., age, gender, region, education, political and religious affiliations etc.) will be included in the survey. The diaries (see below details) provided by members of the citizen science group will be entered through a secure log-in space and will be anonymized (coded) before in-depth analyses.
In the prototyping phase, features, items, and sampling strategies will be developed and consolidated in a digital consensus process involving the project team, expert council, and citizen science group.
The following key dimensions will be considered for the survey:
1. Cognitive understanding (“Corona quiz”): a) Knowledge test (e.g., understanding of risks) and self-assessment, b) Information sources
2. Emotional state (“Corona barometer”): This part will be elaborated with input from the psychology expertise available in the expert council and beyond; data collection will need to be limited to a few indicative key items and cannot replace in-depth psychological studies. Themes will include anxiety/anger/frustration, trust, perceived solidarity vs. exclusion, stigma and discrimination, social isolation vs. resilience.
3. Behavior/behavioral dispositions: a) Compliance with current and past policies (self-assessment; perception of others’ behavior), identify correlations with socio-demographics, information level and emotional state; b) Readiness to comply with potential future public health measures (e.g., exploring acceptance of policies adopted in other countries, such as Italy, Singapore, Taiwan or South Korea), to understand the extent to which certain policy measures would be acceptable to the Swiss public.
4. Moral preferences: These preferences will be probed using the Moral Machine approach, a gamified methodology that has been used for eliciting moral preferences. There is no suggestion that these preferences should be translated into policy, but it opens an opportunity to hear the voices of those who may be affected by policy decisions. Current candidate topics: What restrictive measures are participants willing to accept for what benefit (subgroup analyses), surveillance (identification, contact tracing, monitoring, location tracking etc.), involuntary testing (public health benefit vs. privacy), vaccination (e.g., prioritization of access, incentivized or even obligatory programs etc.), COVID vs. other health care issues.
Space for free text will invite input on additional considerations such as the impact of the pandemic on other health issues, and any other observations participants may consider relevant. Regarding quantitative analysis, automated descriptive and inferential statistics will be performed on selected subscales. Of particular interest will be correlation analyses e.g., between information level and emotional state, and predictive models of adherence to specific public health measures. Advanced analytics will be employed in consultation with the expert group. Selected real-time analyses will be visible for platform visitors upon participation. Real-time subscale mapping will allow for data visualization on a georeferenced overlay. Participants will also get feedback on their knowledge test and a summary of their answers. As a future perspective, participants will get information tailored to their needs and knowledge level.
The full results of the survey will be displayed in PubliCo’s backend, which could in the future serve as a dashboard to policy-makers. The interface will allow for the definition of complex and flexible queries, including factors like time of the answer, geographical and demographic information, and subscale scores. This way it will be possible to find out, for instance, the behavioural dispositions regarding quarantine of the elderly with low anxiety levels living in Appenzell. Localizing queries in time will also allow for a real-time evaluation of the impact of new information or public health measures. The platform is conceived as a quickly adaptable learning system. Its frontend could be enriched with artistic reflections, for instance, by IBME artists-in-residence.
The platform also includes a qualitative component. A selected group of citizen scientists representing different population groups in Switzerland will be engaged in writing bi-weekly open-format diary entries that can be uploaded electronically to the platform or sent as texts or visual material on paper in order to be digitalized by the research team. The heterogeneous group of citizen scientists will be composed of 30-50 participants following a two-step recruitment (purposive sampling extended by snowball sampling with a view to maximum variation). Participants will be given a brief guide to the diary method, emphasizing the openness of the method (e.g., without concerns about spelling and grammar). The guide will ask them to jot down their everyday worries, emotions, risks, (current and past) experiences, decisions and actions in a bi-weekly rhythm, which will allow us “to document changes in values, attitudes, knowledge and behavior”. The diary narratives will be anonymized and analyzed in conjunction with the ongoing data collection by means of thematic analysis, using the software MAXQDA. Various types of text analytics techniques will be explored to add value to manual analysis, for example by searching common topics, by estimating the emotional attitudes of the writer, by identifying items of particular relevance in the diary, or by exploring time trends in common topics. Consolidated NLP techniques will be used to aid in this analysis, in particular topic modeling, sentiment analysis, and named entity recognition. All these automated techniques will be used in a semi-supervised manner, whereas human experts will revise the automated annotation, leveraging them to provide human quality annotations at a faster rate, and at the same time producing training material that will help to further improve the automatic algorithms. The group will experiment with recent language models based on word embeddings and deep neural networks.
Lastly, the development, refinement and in particular the evaluation of the platform and its tools will be supported by focus groups of prospective users (citizens, policy-makers, media representatives, etc.). The focus groups will be conducted and recorded online, using software such as Microsoft Teams. Focus group results will inform the online consensus conferences (oCC), involving the project team, the expert council, and the citizen scientist group, where issues regarding methodology and the interpretation of the empirical findings will be discussed and settled. The oCCs will also discuss normative issues that come up over the course of the project and will be working towards a jointly developed framework and recommendations for best practice standards for RCC and containment measures. The core insights from the oCCs will be presented in PubliCo policy briefs, which will be issued as online publications on a regular (monthly to quarterly) basis.
1. Establish PubliCo – an experimental, interactive online platform collecting data on public perception of COVID-19 and its broader implications (including psychosocial issues and care for non-COVID health conditions), as a feedback loop for policy-makers, health authorities, experts and media professionals engaged in devising and communicating public health strategies for the Swiss public.
2. Design and apply custom-made toolkits (in the form of curated surveys and diary entries) to assess cognitive understanding, emotional state, behavioral dispositions, changes in social practices, and moral preferences of the public (e.g., regarding digital or personal contact tracing, priority access to a future new vaccine or social distancing of high-risk groups).
3. Develop an ethics framework and recommendations for communication, citizen participation and public health interventions, including criteria for “good” communication, preconditions for trust and cooperation, as well as the role of public preferences in relation to public health policy-making and crisis management.
PubliCo Survey will allow to analyze and compare behavior, reactions and adaptation to the COVID-19 crisis across Europe while providing personalized information to the users.
Qualitative interviews will deepen and complement the analysis of quantitative data.
PubliCo will allow policymakers to gain useful insights through advanced queries on the PubliCo dataset; data will be complemented by critical briefs in order to provide interpretation and contextualization of the data. Topics for detailed analyses will be prioritized in close exchange with policy-makers and health systems experts both to ensure analyses address the most pertinent questions.
PubliCo Survey aims to gather information on how COVID-19 and the subsequent different public health measures deployed by different countries impacted on the personal well-being of the participants, intended in a holistic sense. These data will be analyzed both from a national point of view, studying how different groups, defined by gender, age, residency, household arrangement or other factors reacted to the crisis, and from an international point of view, in order to identify efficient strategies for preparedness purposes.
PubliCo surveys will address not only cognitive understanding of the pandemic but also affective states indicative of mental health (anxiety, stress, depression). The project will invite expert advice from psychologists and mental health professionals to ensure optimal coverage. The qualitative component of the study (interviews) will allow to reach elders, people not used to technology, migrants, and other groups that could be underrepresented when employing a digital approach.